Abstract: The basic operators used in Preisach model is hard to be used to describe the dynamics of the hysteresis depending on the input change-rate. A novel tangent function hysteresis operator is, whose shape parameters are related to the input rate, so the shape of the new operator is variable. A dynamic Preisach model is built based on the new operator, and Preisach model parameter are identified by the neural network. Finally, experimental investigation of giant magnetostrictive actuator hysteresis effect is conducted, in this test item, displacement output of giant magnetostrictive actuator excited by sinusoidal current such as 40 Hz, 70 Hz, 100 Hz are sampled as training data, at the same time, ones such as 20 Hz, 50 Hz, 80 Hz and 120 Hz, 130 Hz, 150 Hz are taken as inspection data, experimental results are illustrated to show that the new dynamic model can describe the dynamic hysteresis of the giant magnetostrictive acturtors accurately the effectiveness of the proposed approach. Under multi-frequency signals , the maximum value of the mean square error predicted by this new model is no more than 2.87, and the max absolute error of the new model is no more than 4.4 μm. The research results should be attributed to improvement of accuracy of real-time control systems drived by giant magnetostrictive actuator.

Key words: dynamic Preisach model;giant magnetostrictive actuator;tangent function hysteresis operator